Your test sounds very similar to a granular filter analog of the Gradient Ratio Test (ASTM D 5101), a standardized test method to determine the compatibility of geotextile filters with base soils. The GR test is run in a rigid permeameter using manometers placed at various elevations in the sample to determine head loss within the base soil, and head loss across the filter. There are some modified (better, I think) versions of the test--see literature from Fannin (mid-90s to present) and Fischer (mid- to late 90s). The test allows direct measurement of base soil mass loss.
A highly modified version, called the Flexible Wall Gradient Ratio (FWGR) test was developed by Harney and Holtz in the late 90s (see paper in 2001 North American Geosynth Conf, Portland OR). The FWGR uses a flexible membrane instead of a rigid cell, and uses differential pressure transducers to measure head loss. Note that the differential pressure transducers must be sensitive enough to capture very small head losses (specimen length is ~small). You can determine the proper sensor range and sensitivity by a quick (rough) calc of the head losses you expect for your type(s) of material and your designed length between measurements. The FWGR uses 4 measurement points in the middle portion of the sample - 2 pairs on opposite sides, for redundancy. A fifth measurement is made just upstream of the filter. A sixth measurement is sometimes made just downstream of the upstream edge of the base soil, to check for blinding at this interface (it's been common for these tests to blind at this location, esp if the influent water has not been thoroughly filtered (1-5 micron) and freed of organisms (algae, bacteria, esp if the test will be run long-term).
Sounds like you could use some concepts from these tests, and modify them for your setup.
